Most people think NVIDIA is just a producer of computer graphics cards. However, NVIDIA also makes artificial intelligence (AI) programs, robots, and self-driving cars, and soon the company might be synonymous with public safety, all because of its solution system NVIDIA Metropolis.
Yesterday, NVIDIA proposed the creation of an”AI City” using NVIDIA Metropolis. The premise is quite simple: by 2020, countless video cameras will employ NVIDIA’s Jetson TX2 Module. Cameras will record videos and analyze them with a powerful AI capable of “deep learning” — a highly complex machine learning algorithm that, to my understanding, mimics neural networks and lets the AI learn on its own. The cameras will then send data to data centers/cloud networks that use the NVIDIA Tesla. Ideally, this data will be used for multiple applications, such as helping search for missing children or pets, providing alternate travel routes for GPS systems in real time, and reporting traffic accidents.
According to NVIDIA, cameras without the Jetson TX2 Module can still participate in NVIDIA Metropolis. While the proposed AI cameras can analyze videos on the spot, thus letting them save on bandwidth by only transmitting metadata, regular cameras can still send videos to NVIDIA Jetson or Tesla-powered devices for deep analysis. NVIDIA claims that its GPU-powered system is twenty times more efficient than similar CPU-powered systems.
NVIDIA has already partnered with numerous companies to make NVIDIA Metropolis a reality. These companies include:
- Aeryon (makes unmanned camera drones)
- Aqueti (makes the Mantis Camera, a 100 megapixel camera that has fifty times the resolution of an HD camera)
- Avigilon (develops easy-to-use remote viewing software and HD security cameras)
- BriefCam (creates a video synopsis technology that reviews videos and can index events by time)
- Dahua (makes security cameras that can detect and recognize people and cars)
- Hikvision (makes security cameras that can detect and recognize people and cars)
- Milestone (develops GPU-powered video management software)
- MotionDSP (develops programs that analyze videos for forensic evidence)
- Motionloft (creates real-time person and vehicle-tracking sofware)
- Netradyne (makes vision-based hardware that aids driver performance)
- Robotic Assistive Devices (develops autonomous robots that aid in public safety and monitoring)
- Sensetime (creates face, object, and attribute recognition technology)
- VIMOC (creates software that captures and processes sensory data)
The NVIDIA Metropolis system has the potential to improve public safety. Only time will tell if people accept it.